The Recognition Of Human Face Model

Face Recognition is an important problem in the field of pattern recognition and computer vision. At first, some common methods to recognize faces are presented. The first one is that LDA and feature faces are used to recognize faces at the time. We get the best presenting feature with feature faces, and then get the classifying feature. The second approach to face recognition is kernel independent component analysis (KICA), that is based on an entire function space of nonlinear subspace. The third one is an automatic face recognition method based on support vector machine (SVM), which applies the method of K-L transformation to extract the features of face images and then put these features into SVM for recognition.In this paper, A novel method is proposed to recognize faces in different pose. First, face images are changed into another same face image by the planer projective transformation, then they are represented and classified by the Principal Component Analysis(PCA). Compared with methods that only exploit the Principal Component Analysis(PCA) with the affine transformation, our approach recognize images that are taken from different pose, and classify them with more accuracy. By this way, the unwanted variations resulting from changes in lighting and pose may be eliminated or reduced. We compared the method proposed in this paper with others, to recognize the image in different angles: 0°,6°,12°,18°,24°,30°, which have more effect on other methods, and less on the method proposed in this paper. And this effect occludes their applications in the real world.Theoretical analysis shows that the proposed method can also be more effective in computation. The evaluations on the real images have shown that our methods can provide a better representation and achieve lower error rates in face recognition. Genetic algorithm is also applied to select the number of features that should be extracted in the recognition in order to achieve a higher recognition rate with fewer features.